VM performance - VMware Vs HYPER-V

Hi folks
would be interesting to compare performance of a VM running under VMWare (workstation / player), Virtual Box or HYPER-V.

Using a W7 VM (with latest updates applied) I got a reasonable performance spec on an 8GB HP envy laptop -- I chose W7 as the built in assessement test gives a comparative measurement platform which can be used on any Host -- Max score (i.e 100%) is 7.9


Here's my one on Vmware workstation



If anybody is running a W7 VM on any other Vm software would be good to to see their results. -- I'll see also if I can set up the identical VM on KVM on the same physical hardware.

Cheers
jimbo

Same hypervisor on different hardware gives completely different results so comparing different hypervisors using this method makes no sense.

jimbo45 said:

would be interesting to compare performance of a VM running under VMWare (workstation / player), Virtual Box or HYPER-V...

Then you'll have to choose a different measure of performance. It seems the Windows Experience Index cannot be calculated in a Type 1 Hypervisor like Hyper-V, only in Type 2 like VMWare.

Bree said:

Then you'll have to choose a different measure of performance. It seems the Windows Experience Index cannot be calculated in a Type 1 Hypervisor like Hyper-V, only in Type 2 like VMWare.

Hi there
@Bree

actually seems logical I suppose since VMWare and other type 2 hyper-visors etc don't use "para- virtualised hardware" but provide Virtual hardware so the VM thinks it's running on Native hardware and can address the (what it thinks) is the actual hardware --i.e the "VM Mobo".

I didn't think of that - but thanks anyway for trying it.

Cheers
jimbo

It seems the Windows Experience Index cannot be calculated in a Type 1 Hypervisor like Hyper-V, only in Type 2 like VMWare.

MS always has to have something special... Winsat runs without problems in VM under (even nested, that is why so low score calculated) esxi, which definitely *is* type 1 hv.

Hi there
using a Type 2 Hypervisor obviously has all to carry all the overhead of the HOST OS (running the software) and then use "paravirtualised" hardware etc -- although some limited pass through e.g RAW HDD's allow use of the real hardware by the VM.

The trick is to have as small an OS as possible when using a type 2 VM software and in general you'll probably need a GUI (also overhead when running VM's) and with Windows (not the server editons) being a single user system means probably when that user logs off the VM software (and hence the VM's) aren't available. Fine for desktops etc and software testing.

For 24/7 operation type 1 is obviously the way to go -- my problem is that any sort of VM I've tried running using HYPER-V runs like an absolute dog compared with KVM (mega fast but you do need a lot of expertise in Linux to set it up) or Esxi which is perfect for that type of job -- the whole OS can boot easily from an internal SD card / USB drive if you don't mind a slightly longer boot than by having it from a hard disk --bog problem though for Esxi is that the boot device takes the WHOLE device and can't be run for a partition and that the remaining space can't be used for adding to the data store etc so inacessible for VM's etc.

VMware certainly these days performs really well especially with fast I/O and easy peripheral connections so for home users IMO that's the easiest way to get sensible VM performance and works straight OOB for most people.

Perhaps the next devlopment would be some type of HYBRID possibly type 1.5 HYPERVISOR where some para virtualisation is still present but a lot more (and easier) pass thru to native hardware so even things like extreme gaming can be done seamlessly.

VMWare's proposed rel 16 of VMWare workstation (hopefully out of the "Vaporware stage") seems to be going that way when it announced the intention of allowing it to run alongside HYPER-V --interesting to see how they will implement that one.

anyway "Happy VM-ing" over the holidays.

BTW OT here but I was in Leeds (Yorkshire England) for a few days last week -- I loved the sarcastic humor -- especially some shops displaying as an antidote to the bonkers Black Friday mania -- "Buy Nowt day" !!!!! (Buy Nothing day !!).

Cheers
jimbo

All my gods, stay with me...

bog problem though for Esxi is that the boot device takes the WHOLE device and can't be run for a partition and that the remaining space can't be used for adding to the data store

So wtf is the last partition of 1st disk?? Really enjoying your explanation.





Better to think before writing.

Hi there

@muchomurka

it seemed to want to format THE WHOLE HDD (3.7 TB) when installing the OS -- I checked and re-checked but I couldn't see any way to install esxi to a partition on the HDD and then have the rest of the HDD available as a data space.

The install (on a physical machine) just gave options to install on -- sd card (internal), USB device or selection of the HDD's.

You can have as many partitions as you want on the HDD but when you come to INSTALL the OS you get presented with the available devices (whole devices) not the partitions. -At least that is the choice I got presented with.

At least that was my experience -- if you can show how to install to a partition on a REAL machine I would really appreciate that. Using Esxi itself as a Virtual machine is a whole different ballgame of course.

Cheers
jimbo